Computer and electronic product manufacturing

The pre-screen called this a scale and supply-chain game, which is true of the mainstream and slightly misses why. Assembly is not the barrier: 130 of Canada's 195 establishments have fewer than ten people, and anyone can buy boards, processors, memory and enclosures and ship a working computer. The cut is what is left for the company whose name is on the box. HP buys PC components in volumes no entrant can approach. Its Personal Systems segment took US$38.5B of revenue in fiscal 2025 and kept US$2.05B of it as operating earnings — 5.3% [A]. In the same company, in the fourth quarter, Printing earned 18.9% [A]; the difference is that printing has proprietary supplies and the PC does not. In a computer the parts that determine performance and price — processor, graphics, memory, operating system — are each made by a supplier with more pricing power than the assembler, and they are the same parts in every competitor's machine. Alongside those results HP announced 4,000–6,000 job cuts company-wide [A]. An entrant buys those parts through distribution at worse prices and lower priority than HP does, and has nothing of its own in the product to charge for. What survives at small scale in Canada is, on this reading, the low-volume engineered end — rugged, industrial and embedded computers, custom workstations and servers — where the customer is paying for integration, certification for a harsh environment and long-term availability of an unchanged configuration. Those are real businesses, but each is won one design at a time against Taiwanese industrial-PC makers with the same offer, and none owns the components. The design software sold to hardware companies is screened separately.

Canada has real standing here, and it is not a factory story: the product is engineering, and board assembly can be contracted out. Evertz Technologies of Burlington, Ontario, supplies the routing, processing and playout equipment inside television networks worldwide and reported fiscal 2026 revenue of C$515.8M at a 59.3% gross margin [A]. Those are software-like margins on hardware, and the group's shape — 130 of 246 establishments under ten people, but 23 with a hundred or more — says small design houses can exist. The cut is what it costs to have a product a network or carrier will buy. Evertz spent C$148.1M on research and development in the year — 28.7% of revenue, before C$17.0M of investment tax credits [A], after 29.3% the year before, to keep a catalogue current across every format change its customers go through. A broadcaster or carrier qualifies equipment for a facility it cannot take off air; it buys from the vendor whose gear already runs there and interoperates with the rest of the plant, and radio products must be type-approved country by country before a unit ships. The engineering is sunk before the first order, and then re-spent each technology cycle. And when the engineering stops earning a premium, the hardware is worth little: Semtech agreed in August 2022 to buy Richmond, B.C.'s Sierra Wireless at an enterprise value of about US$1.2B, and in August 2026 agreed to sell the cellular module business that came with it for US$62M [A] — a sale it says sharpens its focus on data-centre and LoRa connectivity. A module is this industry's product with the differentiation worn off. An entrant needs Evertz's spending to get in and risks Sierra's outcome if it cannot sustain it. The design software sold to these companies is screened separately.

Televisions, soundbars and headphones are made by global electronics groups and are not a Canadian entry path; the pre-screen's "brand-and-scale" is right about them. What Canada actually has in this code is hi-fi — loudspeaker and amplifier makers in Ontario and Quebec with international reputations — and that is the proposition screened. It is reachable: 52 of 77 establishments have fewer than ten people, US establishments average 19, and a loudspeaker is drivers, a crossover and a cabinet. The cut is what the category's best brands turned out to be worth. In February 2022 Masimo agreed to pay about US$1.025B for Sound United — eight audio brands, among them Bowers & Wilkins, Denon, Marantz and Polk Audio [A]. In May 2025 it agreed to sell the same business to Samsung's Harman for US$350M in cash [A]. Roughly two-thirds of the price went in three years, for brands with decades of reviews and dealer relationships behind them — sold after a formal strategic review, not out of insolvency. The reading here is that the specialist audio buyer is not being replaced: listening has moved to phones, earbuds and television soundbars sold by platform companies at prices that do not need to earn a return on the speaker, and the specialty dealers who demonstrate and sell separates are fewer each year. An entrant is therefore not fighting incumbents for a growing pool; it is asking for a share of a shrinking one from owners — Harman now holds AKG, Harman Kardon, JBL, Mark Levinson and Revel alongside the Sound United names — who can fill a dealer's whole floor. A craft business selling to enthusiasts is possible; it is a livelihood, not a market entry. The design software sold to electronics makers is screened separately.

Canada has 640 establishments in this code and it is not one market: it runs from navigation and guidance electronics through laboratory and industrial process instruments to the medical devices Statistics Canada files alongside them. What they share is the reason a better sensor is not a business. Nothing here is bought on the instrument alone. A device used to buy or sell anything by measure in Canada cannot be sold into trade until its design is approved — Measurement Canada’s own wording is that "You must obtain a type approval to confirm that a particular measuring device complies with all applicable legal requirements and can be marked for use in trade" [A] — and the medical instruments in this group face the same pattern under Health Canada. Approval attaches to the design, so each variant re-enters the queue, and the cost is sunk before a unit ships. Above that sits the engineering. Teledyne Technologies spent US$317.3M on research and development in 2025 on US$6,115.4M of sales [A], and its Instrumentation segment earned US$400.4M on US$1,457.1M — a 27.5% operating margin [A]. That margin is the point rather than a complaint about it: incumbents here are well paid, and what they are paid for is a qualification an entrant has to buy twice — once from the regulator, and again from a customer who will not put an unproven instrument into a process or a patient it cannot stop. 408 of the 640 Canadian establishments have fewer than twenty employees, so small firms plainly survive; each lives on one or two approved product lines sold to customers it already knows. Entering means funding a line all the way to approval with nothing coming in, and then doing it again for the next variant. Which niche inside the group is most reachable — process instruments, marine, or a medical device — was not tested here and is where a study would start.

The pre-screen called media reproduction a declining category, and for most of this code that is simply right: CDs, DVDs, tape and shrink-wrapped software have gone to downloads and streaming, and what remains is 29 Canadian establishments, 20 of them with one to four people, and one large plant. But Statistics Canada's examples for this code include phonograph record manufacturing, and vinyl is the one format here that is growing, so that is the proposition tested rather than assumed away. The RIAA's year-end report puts US vinyl at US$1,042.9M wholesale in 2025, up 9.3%, on 46.8 million units — a nineteenth consecutive year of growth — while CDs fell 7.8% to US$312.4M [A]. The cut is the size of the pool. That US$1.04B is what labels receive for finished records; the pressing plant is paid a manufacturing fee per disc out of it, and the label keeps the rest. The whole US market is 46.8 million discs a year across every plant on the continent and the European plants that also serve it, and all physical formats together are US$1,381.0M of US$11,535.3M — 12% of recorded music [A]. Growth of 7.9% in units is about 3.4 million more records a year for the entire industry to share, and the orders that fill a plant come from a handful of major-label buyers who place them where they already have capacity and credit terms. What is left for a new plant is short runs for independent artists: real, loyal, and a small business by construction. This screen did not source pressing prices or plant capacity, so it cannot say how small — a study would start there. The design software sold to electronics makers is screened separately.